Internet SEO

Internet Summaries

Google states that page load speed is extremely important, and may affect your ranking. Here is a comparison of some popular website average page-load times as reported by Alexa.





To compare the Alexa average numbers with the Google average numbers, two of my sites are included, with BOTH Alexa and Google numbers, to show the differences.

Column 1 gives the website name. This data was taken on September 6, 2011.
     All Alexa data is taken from http://www.alexa.com/siteinfo/sitename.xxx, i.e.,
     http://www.alexa.com/siteinfo/costco.com

Column 2 is the load time in seconds given by Alexa

Column 3 shows Alexa comments - For fast sites, Alexa says "xx% of sites are Slower -
     For slow sites, Alexa says "xx% of sites are Faster.
     Alexa chooses not to be consistant.
     Alexa actual data is highlighted in yellow in Column 3.

Column 4 shows the "speed class" which corresponds to the Alexa load time given.
     Lines are also drawn comparing my sites' data from Alexa, with my sites' data from Google.
     This illustrates the considerable discrepancy between Alexa and Google.

Column 5 is the load time in seconds given by Google WebMaster Tools.
     MY actual data is highlighted in yellow in Columns 5, 6 and 7 - other numbers are extrapolated.

Column 6 and 7 show "faster than" percentages for each Google load time.

Column 8 show the "speed class" which corresponds to the Google load time given.

Column 9 shows the price paid in load time for using Javascript "look-ups".
Take note how each Google Javascript look-up adds approximately 1/2 second to load time.
It is fascinating that Google Search advises to REMOVE unnecessary Javascript, which they themselves provide. HUH?
To please the "speed folks" at Google, one must DROP all Javascript.
To please EVERYONE ELSE at Google, one should INCLUDE all Javascript. Watcha gonna do?

Why the big discrepancy between Alexa and Google? Some say because Google INCLUDES the Javascript execution, but Alexa does not.
The difference is relatively HUGE - about 1 second.
BOTH Alexa and Google place the dividing line between fast and slow at about 1.6 to 1.7 seconds.
Alexa also provides 3 other speed classifications - average, very fast, and very slow.

Site 1 (slow - 2.6 seconds) has Google Analytics, Google AdSense and Google +1 buttons.
Site 2 (fast - 1.2 seconds) does NOT have Google Analytics, Google AdSense or Google +1 buttons.
Both are on the same server, at the same host.
My conclusions -
     1. Every Java code occurrence adds about 1/2 second to load time.
     2. The Google +1 buttons MUST be removed, to speed up your pages.
     3. Google Analytics SHOULD be removed, to speed up your pages.
     4. Google AdSense will remain (unless speed is your absolute objective).

It would be fantastic if the Google "speed-folks" would talk to their peers, and come up with guidelines.
Since anything slower than 1.6 seconds is officially "SLOW" (WMT), will we get penalized beyond 1.6 seconds?
Or would Google rather just keep us guessing? It is hard to comply with nebulous hints which specify nothing.

You want fast? Please define fast. Google WMT defines fast as "faster than 1.5 seconds".
If that is taken literally, it DEMANDS the removal of ALL Javascript, period.
According to Google WMT, only 20% of sites qualify as "fast".
Google may wish to re-think their various pronouncements, and come up with an INTEGRATED, company-wide policy.
Something like, "1.5 seconds, UNLESS you are using AdSense". "And/or Analytics". "And/or +1".

ALL we want to know is, "When does ANY type of penalty kick-in? ONLY THEN, can we address the issue.
Otherwise, webmasters are going to be DROPPING GOOGLE functions, perhaps needlessly.

Disclaimer
This page consists of a compilation of public information, commonly available on the Internet, at multiple sites, as well as public webmaster forums, and simple manual tests.

Although the author makes every effort to verify the information on this page, no information on this page is guaranteed to be correct, and any data contained herein may be erroneous.

top of page